Comparing Breakthrough and Non-Breakthrough Papers from Early Citing Structures.

Chao Min,Yi Bu,Jianjun Sun

2019-01-01

Abstract:Breakthrough research plays an essential role in the advancement of science system. The identification and recognition of scientific breakthroughs is thus of extreme importance. We propose a citing-structure perspective for observing the unfolding of breakthrough research in the variations of knowledge structure. A series of network topology indictors are used to differentiate the citing networks of over 100 pairs of breakthrough papers and their control papers. 330 pairs of citing networks are subject to statistical tests for those indicators. The results show that compared with less ground-breaking papers, breakthrough papers show salient performance in terms of number of nodes, number of edges, average clustering coefficient, number of components, and average degree. Indicators such as network density, maximum betweenness centrality, and maximum closeness centrality, however, do not have a significantly discriminative power. It reveals that breakthrough papers have more connected citing networks than papers with less ground-breaking ideas. This characteristic has potential implications for early identification of scientific breakthroughs.